08 June 2009, 22:49

We will be evaluating Mahara this year and realise it does not support Oracle, which is our corporate database platform.

For now we will test Mahara using MySQL, however we need to consider a production implementation. It seems our choices will be to change our database platform to suit Mahara (i.e. support Postgres/MySQL), or change Mahara to suit our platform (i.e. Oracle).

Is anyone in the process of including Oracle support in Mahara?  If we decide to go that way we may be able to assist in that effort, or leverage their knowledge to get a head start on our own project.

Either way it would be good to know of others considering this.

09 June 2009, 21:01

Hi - for the record, I know nothing about any efforts underway to make Mahara run on Oracle.

Also, there are a few barriers in the way to Oracle support, which mean that I would recommend if at all possible you should use Postgres instead of trying to spend effort on Oracle support (I wouldn't even seriously consider MySQL, see this post for more information about that).

At this time, none of the development team have experience with Oracle. If we were to support another database platform long term, we'd therefore need someone to maintain this support. It's quite possible some people in the community would be willing and able to help with this, though in the short to medium term this means all the burden of maintenance would have to go on them.

Even if there were people available, someone would have to fix up our database abstraction to support Oracle again. While we grabbed it from Moodle initially, which means this could be done, it's a fair amount of work to do. And to be honest, I don't like the current abstraction. We forked it from Moodle long ago, and it's fair to say it's got its share of bugs as it is. So that job might not be so simple.

[As an aside: I think long term, the solution is to port Mahara to use PHP5's standard PDO - this would make Oracle support a lot easier to implement, and remove a bunch of dodgy code.]

And finally, even if all that effort goes in, the Postgres support is always going to be more stable, and faster.  There are actually some bad bugs in the PHP Oracle driver (the non-PDO one for sure, I don't know about the PDO one) that affect speed for PHP apps using Oracle, that have bitten Catalyst's Moodle clients on Oracle before.

So, in summary: I think the effort involved in you guys deploying a linux server with postgres (or using an existing one, depending on your setup), is going to be much cheaper than adding Oracle support - which will never behave or perform as well as the postgres support.

But hey - if many people are interested and prepared to share costs and expertise, maybe this equation changes